mlw-lt-track-ISSUE-35 (named-entity-syntax-could-use-rdfa-and-microdata): Named entity syntax could use microdata and RDFa in HTML5, and a dedicated syntax in XML only [MLW-LT Standard Draft]

mlw-lt-track-ISSUE-35 (named-entity-syntax-could-use-rdfa-and-microdata): Named entity syntax could use microdata and RDFa in HTML5, and a dedicated syntax in XML only  [MLW-LT Standard Draft]

http://www.w3.org/International/multilingualweb/lt/track/issues/35

Raised by: Felix Sasaki
On product: MLW-LT Standard Draft

Hi Tadej esp. and all,

today I looked at some automatic annotation output I got from Michael. It was created by Enrycher. As I understand it this is experimental, but I wanted to bring one aspect to your attention. The below is simplified:

Input:

            <p>After a century of near domination from the likes of Italy and Germany, international soccer is entering the era of the Cinderella. Russia's Yuri Zhirkov ...</p>
            
Output:

            <p>After a century of near domination from the likes of <span itsx-lexicalizes="dbr:Italy" itsx-entity-type="http://schema.org/Place">Italy</span> and <span itsx-lexicalizes="dbr:Germany" itsx-entity-type="http://schema.org/Place">Germany</span>, international soccer is entering the era of the Cinderella. Russia's <span itsx-lexicalizes="dbr:Yuri_Zhirkov" itsx-entity-type="http://schema.org/Person">Yuri Zhirkov</span> ...</p>


It strikes me that we are probably re-inventing the wheel: large parts of the web community are now heading towards RDFa (light) and microdata for named entities, and we are inventing a new syntax.

So I am wondering whether we shouldn't just describe a best practice to create something like this out of an automatic annotation process:            

            <p>After a century of near domination from the likes of <span itemscope='' itemtype="http://schema.org/Place" itemprop="name">Italy</span> and <span itemscope='' itemtype="http://schema.org/Place" itemprop="name">Germany</span>, international soccer is entering the era of the Cinderella. Russia's <span itemscope='' itemtype="http://schema.org/Person" itemprop="name">Yuri Zhirkov</span> ...</p>    
           

For this, we then already can expect uptake from search engines, and lot's of tools http://schema.rdfs.org/tools.html

I still see a use case for a dedicated "named entity" data category, but rather in a localization chain and in XML, in a workflow like this:

1) HTML is enriched with the microdata result described above, or its RDFa 1.1. light counterpart. 

2) We specify dedicated local markup for entities only in XML, e.g. its:entityType

3) To "glue" 1) and 2) together, we when have a mapping rule like

<its:namedEntityRule selector="//*[@itemtype]" entityTypePointer="@itemtype"/>

No. 1) would also help us with our charter issue, btw.

This approach would also relate to

ISSUE-2 microdata mapping, since we won't map for named entities to microdata and RDFa - they would be available as these from the beginning. 
ISSUE-18 dropping RDFa, since: we won't drop it, but actually do it, at least RDFa light 1.1. 
ISSUE-29 ITS and RDF, since we do 

Thoughts?

Received on Thursday, 5 July 2012 21:48:48 UTC